This article was originally posted in the Montreal Gazette.
The cruelty of COVID is that many people recover only to be left with long-term symptoms. What we have taken to calling long COVID seems to affect about 10 per cent of people post infection. Symptoms tend to resolve slowly over months, if not years, but apart from suggesting patience there was not much medicine could offer. Until now.
Last week saw the publication of a pre-specified secondary end point of the COVID-OUT trial. It is the first high-quality blinded randomized trial to show that anything can be done to reduce the incidence of COVID’s long-term persistent symptoms.
The COVID-OUT trial published its main results last summer. It was conceived at the end of 2020 to test if various pre-existing medications could be used to treat COVID-19. The study tested three potential candidates: metformin (a common diabetes medication), fluvoxamine (an antidepressant), and ivermectin (an anti-parasitic drug). All three medications had potential promise at the time, but all three showed no benefit.
Ivermectin would later become a rallying point for people with fringe views despite the lack of evidence that it did anything. Only metformin showed some potential benefit in maybe reducing ER visits, but did not reduce hospitalizations overall. In short, none of these three medications was useful in treating new COVID infections.
But while they may not be helpful in preventing the severe complications of a new infection, metformin at least seems to decrease the long-term complications of long COVID. In the 10 month follow-up of the study, metformin reduced the incidence of long COVID from 10.4 to 6.3 per cent. The other two medications had no effect.
Reducing rates of long COVID by nearly half would have substantial benefits in reducing the global public health burden of COVID-19. But as the study’s accompanying editorial points out, the trial also provides therapeutic validation for long COVID as a concept. A treatment can only work if there is something to treat.
There are still many uncertainties regarding long COVID, the most important being what actually causes it. There is no universal definition, and researchers had to rely on patients to self-report the diagnosis. But this study makes it harder to deny that it exists altogether, as some people unfortunately claim. If metformin can prevent the disease, then the disease must be real.
This is a single study and you would ideally like to see these results replicated by other groups. Applying this study in the real world will also pose a challenge. The study protocol required the medication to be started within seven days of a COVID infection, with a hint toward more benefit if started within the first four. In an era where many people no longer test for COVID and where testing kits are no longer freely available, I worry that promptly initiating treatment will be difficult.
The study was also done specifically in higher risk groups. Participants had to be over 30 and have either overweight or obesity as risk factors. Whether the benefit still applies to people without these risk factors is unclear.
Finally, the trial tested a 14-day course of metformin in newly infected patients. It’s not clear that metformin will do anything for people who already have long COVID symptoms.
There is no firm rule about when you have enough data to change clinical practice. For some people, and I usually include myself in this group, a single study is not enough to change clinical practice. We obviously want to avoid a similar situation to what we saw with hydroxychloroquine where a surge of unjustified popularity led to a medication shortage for people who really needed it.
But given the lack of alternatives, metformin’s general safety profile (as long as you have normal kidney function), and its low cost, I would not be surprised if guidelines are updated to recommend its use. The practical issues with prompt prescribing to appropriate patients remain, but on the bright side we have for the first time a treatment that can potentially prevent long COVID. The data may not be perfect, but we shouldn’t let the perfect become the enemy of the good.